U.S. patent number 7,295,774 [Application Number 11/302,259] was granted by the patent office on 2007-11-13 for performance monitoring for optical links.
This patent grant is currently assigned to Alcatel. Invention is credited to Henning Bulow.
United States Patent |
7,295,774 |
Bulow |
November 13, 2007 |
Performance monitoring for optical links
Abstract
A method for estimating at least one optical link parameter
using a Viterbi equalizer (1) generating averaged values (M) of
signal amplitudes of a distorted optical signal transmitted through
an optical link (2) for a set of decided bit patterns (a, b, c, d,
. . . ), the method comprising the following steps: forming a first
parameter set (S) characteristic of a distorted signal sequence
using the averaged values (M(a), M(b), M(d), M(g)) for a given
sequence (abdg) of bit patterns (a, b, c, d, . . . ), comparing the
first parameter set (S) to a plurality of reference parameter sets
(R1, R2, . . . ) characteristic for reference signal sequences of
the same sequence (abdg) of bit patterns (a, b, c, d, . . . ), each
of the reference parameter sets (R1, R2, . . . ) having a known
value of the at least one optical link parameter, and selecting the
reference parameter set (R) with the closest correlation to the
first parameter set (S), the known value of the at least one
optical link parameter of the selected reference parameter set (R)
being used as an estimate for the at least one optical link
parameter. A computer program product comprising a software or a
hardware implementing the method.
Inventors: |
Bulow; Henning (Kornwestheim,
DE) |
Assignee: |
Alcatel (Paris,
FR)
|
Family
ID: |
34941934 |
Appl.
No.: |
11/302,259 |
Filed: |
December 14, 2005 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20060177220 A1 |
Aug 10, 2006 |
|
Foreign Application Priority Data
|
|
|
|
|
Feb 4, 2005 [EP] |
|
|
05290259 |
|
Current U.S.
Class: |
398/25; 398/158;
398/33 |
Current CPC
Class: |
H04B
10/00 (20130101); H04L 25/03178 (20130101) |
Current International
Class: |
H04B
10/08 (20060101); H04B 10/00 (20060101) |
Field of
Search: |
;398/25-28,33,34,158 |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
1 422 845 |
|
May 2004 |
|
EP |
|
1 494 413 |
|
Jan 2005 |
|
EP |
|
WO 2004/034611 |
|
Apr 2004 |
|
WO |
|
Other References
Agazzi: "Maximum likelihood sequence estimation in the presence of
chromatic and polarization mode dispersion in intensity
modulation/direct detection optical channels" IEEE International
Conference on Communications, vol. 5, Jun. 20, 2004, pp. 2787-2739,
XP010709741. cited by other .
Wrage, Spinnler: "Distortion identification in WDM networks by
analysis of electrical equalizer coefficients" European Conference
on Optical Communication, vol. 4, 2004, pp. 824-825, vol. 4, Kista,
SE. cited by other .
Agazzi: "Maximum-likelihood sequence estimation in dispersive
optical channels" IEEE Journal of Lightwave Technology, vol. 23,
No. 2, Feb. 2005, pp. 749-763, XP002338784. cited by other.
|
Primary Examiner: Singh; Dalzid
Claims
The invention claimed is:
1. A method for estimating at least one optical link parameter
using a Viterbi equalizer generating histograms of signal
amplitudes of a distorted optical signal transmitted through an
optical link for a set of decided bit patterns, the method
comprising the following steps: (a) forming a first parameter set
characteristic of a distorted signal sequence using averaged values
of signal amplitudes derived from the histograms for a given
sequence of bit patterns; (b) comparing the first parameter set to
a plurality of reference parameter sets characteristic for
reference signal sequences of the given sequence of bit patterns,
each of the reference parameter sets having a known value of the at
least one optical link parameter; and (c) selecting the reference
parameter set with a closest correlation to the first parameter
set, the known value of the at least one optical link parameter of
the selected reference parameter set being used as an estimate for
the at least one optical link parameter.
2. The method according to claim 1, wherein steps are carried
through for a second parameter set using further statistical data
about the signal amplitudes derived from the histograms.
3. The method according to claim 2, wherein the further statistical
data comprises a variance of the signal amplitudes.
4. The method according to claim 1, wherein the comparing in is
carried though by a maximum likelihood estimation, in particular by
calculating a minimum square error.
5. The method according to claim 1, wherein the at least one
optical link parameter is selected from the group consisting of
chromatic dispersion, polarization mode dispersion, self-phase
modulation, and optical signal to noise ratio.
6. The method according to claim 1, wherein forming a first
parameter set characteristic is preceded by a reference parameter
set generating step, generating a reference value for each bit
pattern and for each reference parameter set.
7. The method according to claim 1, wherein the sequence of bit
patterns is derived from a sequence of transmitted bits.
8. The method according to claim 1, wherein three subsequent bits
are used to form a bit pattern.
9. A computer readable recording medium having embodied thereon a
computer program for estimating at least one optical link parameter
using a Viterbi equalizer generating histograms of signal
amplitudes of a distorted optical signal transmitted through an
optical link for a set of decided bit patterns, which is programmed
to perform, when executed on a computer, comprising the following
steps: (a) forming a first parameter set characteristic of a
distorted signal sequence using averaged values of signal
amplitudes derived from the histograms for a given sequence of bit
patterns; (b) comparing the first parameter set to a plurality of
reference parameter sets characteristic for reference signal
sequences of the given sequence of bit patterns, each of the
reference parameter sets having a known value of the at least one
optical link parameter; and selecting the reference parameter set
with a closest correlation to the first parameter set, the known
value of the at least one optical link parameter of the selected
reference parameter set being used as an estimate for the at least
one optical link parameter.
Description
The invention is based on a priority application EP 05290259.0
which is hereby incorporated by reference.
FIELD OF THE INVENTION
The invention relates to a method for estimating at least one
optical link parameter.
BACKGROUND OF THE INVENTION
For performance monitoring of 10 Gb/s (future 40 Gb/s) transmission
systems, it is mandatory to obtain a reliable estimate of the
status of optical transmission links for preemptive fault detection
for network management or network control. This status can be
described by optical link parameters being characteristic of the
distortion of an optical signal transmitted through the optical
fiber link, such as chromatic dispersion (CD), polarization mode
dispersion (PMD), self-phase modulation (SPM), etc.
There have been proposed many costly solutions for performance
monitoring of optical links, including optical measurement
techniques as well as techniques which use additional modulation
tones (not in the standard). At the ECOC 2004, M. Wrage and B.
Spinnler have proposed a method of distortion identification by
analyzing equalizer coefficients (FFE tap settings) of a finite
impulse response (FIR) equalizer in the paper "Distortion
Identification in WDM Networks by Analysis of Electrical Equalizer
Coefficients". The solution proposed in this paper is based on the
equaliser setting of a FFE, which is used for all signal bit
patterns. Due to the limited parameter set (taps) it works only for
a limited dispersion range or fails if mixed distortions are
present (e.g. PMD and CD).
OBJECT OF THE INVENTION
It is the object of the invention to obtain a reliable estimate of
the status of an optical transmission link by determining estimates
of optical link parameters being characteristic for distortion
induced in the optical link.
SHORT DESCRIPTION OF THE INVENTION
This object is achieved by a method for estimating at least one
optical link parameter using a Viterbi equalizer generating
averaged values of signal amplitudes of a distorted optical signal
transmitted through an optical link for a set of decided bit
patterns, the method comprising the following steps: (a) forming a
first parameter set characteristic of a distorted signal sequence
using the averaged values for a given sequence of bit patterns, (b)
comparing the first parameter set to a plurality of reference
parameter sets characteristic for reference signal sequences of the
same sequence of bit patterns, each of the reference parameter sets
having a known value of the at least one optical link parameter,
and (c) selecting the reference parameter set with the closest
correlation to the first parameter set, the known value of the at
least one optical link parameter of the selected reference
parameter set being used as an estimate for the at least one
optical link parameter.
The inventive method uses a Viterbi equalizer which already sorts
the received signal (samples of it) according to the bit pattern.
Each bit pattern consists of a sequence of subsequent decided bits
being correlated by inter-symbol interference (ISI), the ISI being
characteristic for the distortion of the optical link. Therefore,
much more parameters as in the state of the art can be used to
quantify and identify different distortions (=dispersions).
For a given sequence of bit patterns, a first parameter set
consisting of a sequence of averaged values characteristic for the
unknown optical link parameter (distortion) is generated. This
parameter set is compared to reference parameter sets consisting
each of a sequence of averaged values with a known value of the
optical link parameter. The reference sequence having the parameter
set which is closest to the first parameter set is chosen as an
estimate for the optical link parameter. It is evident that the
precision with which the optical link parameter is estimated
depends on the number of reference parameter sets used. If two
reference parameter sets have about the same (closest) correlation
to the first parameter set, the estimate of the optical link
parameter is accomplished by interpolating between the known values
of the two reference parameter sets.
In a preferred variant, the averaged values of the signal
amplitudes are derived from histograms of signal amplitudes
generated by the Viterbi equalizer. For enabling the decisions of
Viterbi equalizers (channel modeling) and avoiding erroneous
decisions it is well-known to use a monitoring device generating
statistical data about the signal transmitted through the optical
link. In such a device, a histogram of a probability distribution
of signal amplitudes of an optical signal transmitted through the
optical fiber link is generated. Each peak of the probability
distribution, which is about equal to the mean value, is
characteristic for the received signal amplitude belonging to one
of the bit patterns. It might be appropriate to characterize a bit
pattern by a signal amplitude at or close to the temporal center
position of the bit pattern.
In a further preferred variant, steps (a) to (c) are carried
through for a second parameter set using further statistical data
about the signal amplitudes derived from the histograms. The
histograms may serve to determine more statistical data than just
the mean value for each bit pattern.
In a further variant, the further statistical data comprises the
variance of the signal amplitudes. The mean value of the signal
amplitude may be used in conjunction with the variance of the
signal amplitude for determining certain optical link parameters
such as the optical signal to noise ratio (OSNR).
In a preferred variant, the comparison in step (c) is carried
through by a maximum likelihood estimation, in particular by
calculating a minimum square error. In this way, an easy comparison
of the reference parameter sets with the first parameter set is
provided.
In a highly preferred variant, the at least one optical link
parameter is selected from the group consisting of chromatic
dispersion (CD), polarization mode dispersion (PMD), self-phase
modulation (SPM), and optical signal to noise ratio (OSNR). These
(and other) properties of the optical fiber link may be
advantageously determined by the inventive method.
In a further variant, step (a) is preceded by a reference parameter
set generating step, generating a reference value for each bit
pattern and for each reference parameter set. The reference value
may be the averaged value, the variance, etc. of the signal
amplitude.
In a highly preferred variant, the given sequence of bit patterns
is derived from a sequence of transmitted bits. In this way, an
on-line determination of optical link parameters for a transmitted
signal is possible.
In yet another variant, three subsequent decided bits are used to
form a bit pattern. The three subsequent bits are correlated
through inter-symbol interference (ISI), i.e. the transmitted bit
in the center of the three subsequent bits is influenced by the bit
transmitted directly before and the bit transmitted directly after
the bit in the center. The ISI is due to signal distortion.
The invention is also realized in a computer program product
comprising a software or a hardware implementing the method as
described above. The computer program product may be implemented as
a hardware in a DSP-ASIC or as a software in a processor reading
the histogram data of the Viterbi-ASICs. The dispersion values may
then be made available to the network management or the control
plane of the optical network.
Further advantages can be extracted from the description and the
enclosed drawing. The features mentioned above and below can be
used in accordance with the invention either individually or
collectively in any combination. The embodiments mentioned are not
to be understood as an exhaustive enumeration but rather have
exemplary character for the description of the invention.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention is shown in the drawing.
FIG. 1 shows a Viterbi equalizer used for implementation of the
inventive method and a flow-chart of the inventive method.
DETAILED DESCRIPTION OF THE INVENTION
The technique explained below is based on the existence of a
Viterbi equalizer 1 as shown in FIG. 1. The Viterbi equalizer 1
(maximum-likelihood sequence detector) may be implemented in the
receiver line card of an optical receiver (not shown). For its
adaptation, the Viterbi equalizer 1 extracts detailed information
on the distortion of an optical signal transmitted through an
optical link 2, the optical signal being converted to an analog
electrical signal by a photodiode (not shown) at the entrance of
the Viterbi equalizer 1. The analog electrical signal is converted
to a digital bit sequence in a Viterbi core 3.
Depending on the subsequent bits appearing at the output of the
Viterbi core 3, a shift register 4 distributes the input to the
Viterbi core 3 to one of the output channels of a multiplexer 5.
The outputs are fed into a histogram generator 6. Commonly the
signal at the input of the multiplexer 5 is analog-to-digital
converted (not shown in the FIGURE) and the histogram generator 6
comprises digital registers. However, it is also possible to use an
analog multiplexer as an alternative solution. The process
described above will be explained in greater detail below:
The shift register 4 contains a first, second and third bit ijk and
is connected to the output of the Viterbi core 3. The second bit j
is the decided bit for a given sampling time (t), the first bit i
is the decided bit at a previous sampling time (t-1), and the third
bit k is the decided bit at a subsequent sampling time (t+1). The
contents of the shift register 4 are shifted to the left after each
time step.
The reason why three bits ijk are registered in the shift register
4 is that the correlation length of the inter-symbol interference
(ISI) is one bit, i.e. only the subsequent, third bit k and the
previous, first bit i have an influence on the analog value of the
optical signal measured for the second bit j. There are eight
possible states of the shift register 4, each one defining a bit
pattern. The first bit pattern a is identified with the three-bit
state ijk=000, the second bit pattern b is identified with the
three-bit state ijk=001, etc.
The shift register 4 is connected to the multiplexer 5 whose input
is connected to the input of the Viterbi core 3. The analog value
of the signal at the input of the Viterbi Core 3 is therefore
delivered as an input to the multiplexer 5 which selects one out of
its eight possible output channels in dependence of the state of
the shift register 4. Thus, the analog values for each of the eight
decided bit patterns are delivered in separate channels to the
histogram generator 6.
The histogram generator 6 uses the analog values or the digitized
analog values measured for each of the bit patterns at the entrance
to the Viterbi core 3 to generate a probability density function
pdf(a), pdf(b) etc. for each of the bit patterns (channel model).
An output of the histogram generator 6 is used as a feedback signal
to the Viterbi core 3, adapting the branch metric unit in
dependence of the statistical data generated in the histogram
generator 6. From the probability density functions pdf(a), pdf(b),
. . . the mean values M(a), M(b), . . . of signal amplitudes of the
analog signal can be derived for each bit pattern a, b, etc. The
mean values M(a), M(b), . . . measured in FIG. 1 are obtained from
statistics about the analog value of the signal at the input to the
Viterbi Core 3 at the sampling time of the second bit j.
It is practical to evaluate the signal amplitude only at the
sampling time of the actual transmitted bit j. However, it is also
possible to determine the value of the signal amplitude over a
short time interval before and after the decision time of the
second bit j, the time interval being much shorter than the overall
duration of the three-bit sequence ijk. The time average over this
short time interval may be taken in the histogram generator 6
before the statistical evaluation is carried through.
The statistical information about the optical signal in the
histogram generator 6 is dependent on the distortion of the optical
link 2, as the ISI is distortion-dependent. Consequently, the data
being present in the histogram generator 6 can be used to calculate
optical link parameters such as chromatic dispersion (CD),
polarization mode dispersion (PMD) etc., with the method described
below.
For this method to work, it is necessary that the averaged values
M(a), M(b) etc. for each of the eight bit patterns are determined
in the Viterbi equalizer 1. These values being known, a sequence of
bit patterns, e.g. abdf, is defined in a first step 7 of the method
and a sequence of mean values M(a), M(b), M(d), M(g) is generated
from this sequence, so that a first parameter set S(M(a), M(b),
M(d), M(g)) is formed.
In a second step 8, the first parameter set S is compared to a
plurality of reference parameter sets R. The reference parameter
sets R consist of a sequence of averaged values of the analog
signal with known values of the distortion parameters to be
estimated, in the present case CD and PMD. The first/second
reference parameter set R1/R2 corresponds to a first/second value
CD1/CD2 of chromatic dispersion and a first/second value PMD1/PMD2
of polarization modulation dispersion. Of course, it is possible
for the reference parameter sets R to differ only in one optical
link parameter, such that in the above case, the values of
polarization mode dispersion are equal for all of the reference
sequences R whereas the CD values are not.
In a third step 9, the reference sequence with the closest
correlation to the first parameter set S is chosen. The correlation
is defined in the sense of a maximum-likelihood correlation, being
evaluated e.g. by calculating the minimum square error between the
parameter set S and the parameter sets R, i.e. min(S-R).sup.2. The
reference parameter set with the closest correlation to the first
parameter set S is determined and the corresponding distortion
values CD, PMD define the estimated link parameters for chromatic
dispersion and polarization mode dispersion, respectively.
Because of the use of averaged values a noise-free (but still
distorted) analog sequence is reproduced. This analog sequence may
or may not have been transmitted through the optical link 2. It is
sufficient that the sequence of bit patterns is significant for the
analog signal. However, the sequence of bit patterns it is chosen
preferably in such a way that it has a significance for the
transmitted signal. Especially, it is possible to derive the
sequence of bit patterns from a sequence of decided bits
transmitted through the optical link 2, as is described below.
As an example for the connection between bit sequences and bit
patterns, it is supposed that for a given time step a bit sequence
consisting of three subsequent bits a=000 corresponding to the
first bit pattern a is present in the shift register 4. In the
following time step, a new decided bit "1" is present at the output
of the Viterbi core 3. Consequently, the shift register and adopts
the state ijk=001 corresponding to the second bit pattern b. In the
following time step, a new decided bit "1" is present at the output
of the Viterbi core 3, leading to a state ijk=011 of the shift
register 4, corresponding to the fourth bit pattern d. If the
decided bit of the subsequent time step is "0", the state of the
shift register 4 is ijk=110, corresponding to the seventh bit
pattern g.
In this way, the sequence of decided bits transmitted through the
optical fiber link 2 consisting of six subsequent bits 000110 gives
rise to the sequence of four bit patterns abdg. It is thus possible
to carry through the three steps 7, 8, and 9 with a sequence of bit
patterns being derived from the sequence of transmitted bits. The
length of the sequence of bit patterns may be chosen to be four, as
above, though it is also possible to use a sequence consisting of
more or less bit patterns.
The method described above may be easily modified by using
additional statistical data from the histogram generator 6, e.g.
the variance of the amplitudes of the bit patterns a, b, etc. Thus,
a second, third etc. parameter set can be generated for each
sequence of bit patterns. The correlation of these parameter sets
and the reference parameter sets for these additional parameters
can be calculated as described above. As a result, optical link
parameters with greater accuracy can be obtained. In this way, it
is also possible to estimate the signal-to-noise ratio. The
variances, which can be extracted from the different histograms,
are directly linked to the optical signal to noise ratio (OSNR).
The lower the OSNR, the more noise and the bigger are the
variances. The OSNR values can be determined by comparing the
histogram variances with the variances of reference patterns R with
a known OSNR, or can directly be calculated from the histograms
using theoretical approaches.
The reference parameter sets R can be obtained before the first
step 7 by generating a look-up table in which an averaged value of
the analog signal is stored for each value of the at least one
optical link parameter and for each bit pattern. These averaged
values have either been measured or obtained by numerical
simulation for each distortion parameter separately or for mixed
distortions (i.e. CD and PMD). It is also possible that the
reference parameter sets are calculated on-line in a processor
while performing the method as described above. The method
described above may be implemented as hardware or software as part
of a Viterbi equalizer control.
In summary, when a Viterbi equalizer is present in an optical
receiver, the above method for the determination of distortion
parameters can be easily implemented almost without additional cost
as a software in the Viterbi equalizer. As Viterbi equalizers
generally adapt themselves to the incoming signals in the range of
milliseconds, a millisecond speed for performance monitoring of
optical links can be reached. Furthermore, the method described
above uses more parameters than the state of the art and
consequently has the potential to be more precise.
* * * * *